Absolute single-particle detection efficiencies and pulse height distributions of a cascaded microchannel plate (MCP) detector are reported for the ions H2+, Ne+, Ar+, Kr+ and Xe+ in the impact energy range 0 - 4.75 keV. The detection efficiencies have been obtained using the photoelectron - photoion coincidence (PEPICO) technique which allows the determination of detection efficiencies without knowledge of the absolute ionization rate. The mean gain is found to increase with increasing ion impact energy and decreasing ion mass. Depending on the ion mass, the absolute detection efficiencies saturate above 2.5 keV and reach about 63%.
Microchannel plates (MCPs) are frequently used in experimental research instrumentation systems for the detection of electromagnetic radiation and neutral and charged particles. One important MCP application has been in the detection of positive ions in many designs of mass or energy analysers. The absolute detection efficiencies of MCPs for ions are of particular interest if absolute intensities of particle currents are required or if the counting rates of different ions have to be related to each other. However, information on the detection efficiencies for particles is very limited (Fraser 1989). In a recent work that has been carried out in our laboratory (Brehm et al 1995), the absolute single-particle detection efficiencies of a MCP for ions were determined by using the photoelectron - photoion coincidence (PEPICO) technique (Brehm and von Puttkamer 1967). In the present work, we report on new and extended results for H2+ and rare gas ions obtained with an improved experimental set-up.
The PEPICO detection system has been described previously (Brehm et al 1995); however, some significant improvements have been made. Briefly, a light beam in the vacuum-ultraviolet region is produced by a gas discharge lamp. A monochromator (1 m McPherson) selects the desired wavelength and focuses the light beam onto the target beam effusing from a capillary array. Electrons and ions are produced by photoionization of the target atoms or molecules. A potential difference of 600 V between two parallel grids (numbers 1 and 2) produces a homogeneous electric field across the ionization volume and the charged particles are accelerated in opposite directions depending on the sign of their charge. After passing grid number 1, the electrons are finally accelerated to 670 eV before they strike the front side of a cascaded MCP detector (chevron). After their primary acceleration, the positive ions enter a field-free region between grid number 2 and a third grid which is held at the same potential as grid number 2. The length of the field-free region is twice the distance of the ionization volume to the primary acceleration grid. By this means, time-of-flight focusing is achieved, resulting in a considerable reduction of the time-of-flight widths which are typically 15 ns (full width at half maximum). After passing through the field-free region, the ions are accelerated by means of a fourth grid towards a second cascaded MCP detector (chevron). The ions hit the MCP surface at right angles and with impact energies in the range of 0 - 4.75 keV. The front side of the first MCP is biased at +50 V with respect to the final acceleration grid in order to drive electrons emitted from the interchannel web of the MCP back to the MCP (Brehm et al 1995). This electron back-driving field increases the relative detection efficiency from about 0.75 to 1.0. Each of the grids has a transmission of 78%. The electric charge pulses from the MCP detectors are fired into commonly used coincidental electronics. The electron pulses are used to start a time-to-amplitude converter; the ion pulses provide the stop signal. In this way, time-of-flight mass spectra are obtained. The time-of-flight differences are in the range 0.5 - 4.25 microseconds for the ion mass range 2 - 131 u. For calibration purposes, the ion pulses are preamplified by a charge-sensitive amplifier accounting for a pulse height analysis of the MCP output pulses. The vacuum conditions have also been improved. Among other things, the vessel containing the detection system is now evacuated by a turbomolecular pump that guarantees a vacuum nearly free of oil. Without running the gas beam, the background pressure is below 6*10-8 hPa, whereas with the gas beam the background pressure increases by a factor of 100.
The MCPs used are of type G12-36ST/15/T from Philips Photonics. These plates have 36 mm outer diameter, a channel diameter of 12.5 micrometer , a channel bias angle of 15°, a channel length-to-diameter ratio of 40:1 and an open area of at least 60%. The electrode at the input surface has been deposited half of the channel diameter into the channels whereas at the output surface the electrode penetration is about two channel diameters. The deep electrode deposition at the back side of the MCP is intended to lower the angular spread of the outgoing electrons (the focusing effect) because these MCPs are preferentially used for image-intensifier applications or position-sensitive particle detectors. Each MCP used for the ion detector is operated at 850 V.
The data evaluation procedure has been described
by Brehm et al (1995). In short, the number of true coincidences
is taken from the time-of-flight mass spectra;
and 
denote the effective detection efficiencies
for the ions and electrons, respectively,
and 
represents the total number of photoionization
events. Together with the counting rate for
the electrons 
and the subtraction of accidental coincidences
and background electrons 
, the effective detection efficiency of the
ions can be expressed as 
. The effective detection efficiency is given
by three contributions. (i) The ion collection
probability T depends on the transmission of the three
grids alone 
if both the geometrical arrangement and
the applied electric fields ensure that all
ions are guided to the MCPs. Saturation measurements
as well as trajectory calculations prove
that an electric field strength of 15 kV
across the ionization volume is sufficient
to collect all electrons and ions from the
ionization volume. (ii) The pulse-height
electronics needs discrimination against
background pulses resulting from external
sources and amplifier noise. So, real MCP
pulses of small charge are cut off. From
the analysis of the pulse-height distributions
of the MCPs and the pulse processing of the
discriminator unit, the pulse detection probability
D is obtained. (iii) The absolute detection
efficiency 
of the MCP for ions which is the subject
of the present work. Hence, the effective
detection efficiency is given by 
= 
and the following equation is used for the
actual data evaluation:
It has to be noted that the absolute detection efficiency is given by measured quantities only; knowledge of the absolute event rates is not required. Possible sources of error and their elimination have been discussed by Brehm et al (1995).
Figure 1. The pulse height distributions of the MCP
stack at ion impact energy of 1 keV (left-hand
column) and 4.5 keV (right-hand column).
The abscissa has been calibrated with respect
to the absolute number of output electrons.
The MCP stack has been operated with 850
V per MCP. The vertical lines at 
indicate the setting of the discriminator
level (see text).
Figure 1 shows the pulse height distributions for
H2+, Ne+, Ar+, Kr+ and Xe+ at the two selected impact energies 1 keV
(left-hand column) and 4.5 keV (right-hand
column), respectively. The abscissa has been
calibrated with respect to the absolute number
of output electrons. For each distribution,
the dark-pulse distribution is obtained with
the gas beam off and diffuse gas inlet at
the same residual pressure. This signal is
subtracted from the distributions measured
with the target beam on. The dark count rises
significantly below 
output electrons. In Figure 1, the vertical lines at 
indicate the setting of the discriminator
level. Charge pulses below that level are
cut off and are not detected if time-of-flight
mass spectra are recorded (see above). With
increasing impact energy, the peaked pulse
height distributions become broader. The
maximum position, namely the modal gain of
the MCP stack, and the centre of gravity,
namely the mean gain (mean output charge)
of the MCP stack, are shifted towards higher
charge quantities. A striking observation
is the double-peaked pulse height distribution
for H2+ at an impact energy of 4.5 keV. At present,
there is no confirmed explanation of the
significance of this structure. If real,
this observation requires further study.
In Figure 2, the mean gain is plotted as a function
of the ion impact energy. Since the ion-induced
electron emission yield increases with ion
velocity (Higatsberger et al 1954, Parilis and Kishinevskii 1960, Hellsing
et al 1985), both the modal gain and the mean
gain should be affected by the total number
of initially produced electrons (Brehm et al 1995) which was confirmed by our experiment.
This observation had already been made by
Meier and Eberhardt (1993) when they investigated
the gain of MCPs for different ions. Those
authors successfully described the velocity
dependence of their data in terms of the
Parilis - Kishinevskii relation (Parilis
and Kishinevskii 1960) which approximates
the ion-induced electron emission yield by
where the coefficients a and b are characteristic for each ion species.
For ion velocities v much larger than the threshold velocity
, the relation becomes linear and the electron
yield depends on the impact energy E of an ion with mass m as 
. The full lines in Figure 2 show fits of the Parilis - Kishinevskii
relation to our data. Obviously, this relation
describes the velocity dependence of the
mean gain in a reliable way. For the rare
gas ions, the mass dependence of the mean
gain at fixed impact energies is described
very well by the Parilis - Kishinevskii relation.
Figure 2. The mean gain (mean output charge) of the
MCP stack for the ions under investigation
as a function of the ion impact energy at
an operating voltage of 850 V per MCP. The
full lines show fits of the Parilis - Kishinevskii
relation to our data (see text).
Figure 3. The absolute detection efficiencies of the
MCP as functions of the ion impact energy.
For the rare gas ions under investigation,
the broken line shows the saturation of the
absolute detection efficiencies at 63.5%.
The absolute detection efficiencies 
as functions of the ion impact energies
are shown in Figure 3. The indicated error bars represent two
standard deviations as calculated from the
statistical uncertainty of the counting rates
plus the error in determining the pulse detection
probability D. For the rare gas ions under investigation,
the absolute detection efficiency increases
from a threshold of around 0.5 keV to a more
or less constant detection efficiency of
which is close to the open area of the MCPs.
The maximum absolute detection efficiency
for H2+ has been found to be 
. As pointed out in the previous work (Brehm
et al 1995), the detection efficiency of a MCP
should be ruled essentially by the initial
detection efficiency for the incoming particles.
If the ion-induced electron emission yield
rises with increasing impact velocity and
if it reaches such high values that the probability
for no electron emission is reduced to negligible
values, then the onset of the saturation
is shifted to higher impact energies with
increasing ion mass. This behaviour can be
seen from figure 3. The absolute detection efficiencies are
found to be independent of the total rate
of ionization events given by the light beam
intensity, the target beam density and the
photoionization cross section. Different
operating voltages of the MCPs entail a variation
in the pulse height spectra but no change
in the absolute detection efficiencies. This
observation supports the assumption that
the absolute detection efficiency is mainly
given by the initial detection efficiency
of the incoming particles. Several MCPs of
the same type show identical absolute detection
efficiencies.
The asymptotic values of the absolute detection efficiencies can be understood from trajectory calculations of the initially produced secondary electrons and their yields. In the case of Ar+ ions with 3 keV impact energy, the secondary electron yield for the Nichrome-coated interchannel web is about 2.3 (Higatsberger et al 1954) and that for MCP material is around 5 (Hellsing et al 1985). If we assume for simplicity a Poisson distribution of the number of secondary electrons, then the probabilities for the emission of electrons are close to unity when an ion strikes the interchannel web or the open area. The electron-induced secondary electron yields are considerably lower. From the present operating conditions of the MCP, we estimate an electron impact energy of about 50 eV inside a channel. The same value can be taken for those electrons that are driven back due to the bias voltage between grid number 4 and the front side of the first MCP (see above). No published data have been found for Nichrome; limited data for nickel (Weast 1987) indicate a yield far below 1 for 50 eV electrons which gives a very low secondary electron emission probability. The electron-induced secondary-electron yield from reduced lead glass (MCP material) is about 1.5 (Authinarayanan and Dudding 1976) for 50 eV electrons which gives a high secondary electron emission probability. If no electron back-driving field is applied in front of the MCP, electrons from the interchannel web as well as electrons from the entrances of the channels may be accelerated away from the MPC surface, leading to a decrease in the detection efficiency. With increasing strength of the electron back-driving field more and more of these electrons are driven back to the MCP. Under the present operating conditions, most of these electrons hit the MCP very close to their origin; that is, electrons from the interchannel web strike the interchannel web again and vice versa. Hence, significant contributions to the development of an electron avalanche inside a channel come from ions striking the open area which approximately gives the maximum detection efficiency of the MCP.
When the MCPs are turned over so that the ions hit the back side of the MCP, then the absolute detection efficiencies are reduced to about 45%. This demonstrates the influence of the deeper electrode deposition on the absolute detection efficiency and can be qualitatively understood from the strongly different electron yields. At normal incidence to a 15° biased channel, ions and electrons will penetrate about 3.7 channel diameters. The lead glass fraction of the channels is very substantially decreased by electrode penetration of two channel diameters at the back side of the MCP. Consequently, we observe a substantial decrease in the absolute detection efficiencies. Because of the limited data, a quantitative analysis is hard to give at the present state of the experiments.
A direct comparison of the present results with the few other works on the absolute detection efficiencies of MCPs for ions (Brehm et al 1995) is rendered difficult, since the efficiencies depend on the details of construction of a MCP. However, the present results are in excellent agreement with the work of Gao et al (1984) who determined detection efficiencies of another type of MCP by comparing very low currents of H+, O+ and He+ with the corresponding counting rates produced by the MCP detector.
It would be of great importance to determine the absolute detection efficiencies of various other ions and for other types of MCP used for physical applications. The constructional details of a MCP, for example the materials used for the MCP production, the electrode penetration into the channels, the open area, the channel bias angle or the ion impact angle, play an important role for the absolute detection efficiencies. These technical features may be improved for ion detection if detailed studies are performed. Besides that, the investigation of the distinctive properties of the H2+ pulse height distributions and their absolute detection efficiency (or those of other molecular ions) may give deeper insight into the physical processes operative in a MCP. Further work along these lines is under way and the results will be reported in a forthcoming publication.
We are indebted to Professor Dr B Brehm (Universität Hannover) for his support and many helpful discussions. We would like to thank Philips Photonics for providing us with the microchannel plates used for the present work and we gratefully acknowledge the cooperation with S O Flyckt (Philips Photonics, Brive, France).
References
Authinarayanan A and Dudding R W 1976 Changes in secondary electron yield from reduced lead glasses Adv. Electron. Electron Phys. A 40 167 - 81
Brehm B, Grosser J, Ruscheinski T and Zimmer M 1995 Absolute detection efficiencies of a microchannel plate detector for ions Meas. Sci. Technol. 6 953 - 8 and references therein
Brehm B and von Puttkamer E 1967 Koinzidenzmessung von Photoionen und Photoelektronen bei Methan Z. Naturf. a 22 8 - 10
Frazer G W 1989 X-ray detectors in astronomy (Cambridge: Cambridge University Press) pp 116 - 78
Gao R S, Gibner P S, Newman J H, Smith K A and Stebbings R F 1984 Absolute and angular efficiencies of a microchannel-plate position-sensitive detector Rev. Sci. Instrum. 55 1756 - 9
Hellsing M, Karlsson L, Andrén H-O and Nordén H 1985 Performance of a microchannel plate ion detector in the energy range 3 - 25 keV J. Phys. E: Sci. Instrum. 18 920 - 5
Higatsberger M J, Demorest H L and Nier A O 1954 Secondary emission from Nichrome V, CuBe, and AgMg alloy Targets Due to positive ion bombardment J. Appl. Phys. 25 883 - 6
Meier R and Eberhardt P 1993 Velocity and ion species dependence of the gain of microchannel plates Int. J. Mass Spectrom. Ion Processes 123 19 - 27
Parilis E S and Kishinevskii L M 1960 The theory of ion - electron emission Sov. Phys. - Solid State 3 885 - 91
Weast R C (ed) 1987 CRC Handbook of Chemistry and Physics 68th edn (Boca Raton, FL: CRC)